Particle crushing is closely related to many geotechnical engineering problems and is an interesting topic that has been studied intensively in recent decades. Particle crushing is affected by many factors, e.g., size, heterogeneity, morphology, mineralogy and environmental factors (e.g., erosion, fire and flood), making it a heretofore unsolved problem of great geotechnical importance. The aims of this research are to identify the fracture surfaces of fragments that are resulted from the mechanical crushing of a single sand particle, to investigate the role of flaws in the particle crushing behaviors and to propose a novel framework for a discrete element method (DEM) simulation of mini-triaxial tests on sand with realistically shaped grains and modeling of crushable granular materials during the continuous breakage process in a one-dimensional (1D) compression test.

In the first phase of this research, a fracture region matching algorithm was developed to reassemble fractured sand particles by using a variety of image processing and matching techniques. In the second phase, the target of research was set to match the individual fragments to their mother particle directly, which entailed matching the original face (if any) on the fragment surface to the mother particle and then identifying the fracture face on a fragment surface that was generated from the mechanical crushing event. In the third phase, the research endeavored to enhance the algorithm to realize the child-mother particle matching within a small assembly of crushed particles. Specifically, 62 child particles resulted from single particle crushing tests of 9 Leighton Buzzard sand (LBS) particles would be mixed up and sorted according to their volumes to reach a large-deformation state. As a further step towards the goal of exploring the role of initial flaws in the particle crushing behaviors, the following work endeavored to incorporate the authors’ earlier research output into the DEM model to reproduce the crushing behavior observed in the experiment. In parallel with the above research work, one-to-one modeling, where every particle in the physical experiment has its own numerical twin, and a particle fragmentation scheme (i.e., creating virtual templates of fragments using computed tomography (CT) data of crushed particles) were proposed and applied to a set of published physical test results.

It can be concluded that the proposed method in the first phase is capable of reassembling fractured particles and it would facilitate the better prediction of particle fracture in numerical modeling. 9 LBS particles were successfully reassembled using the techniques mentioned in the second phase, demonstrating the high competence and robustness of the techniques in quantifying the fragment morphologies and matching them to the original particles. The major advancement in the third phase is the capability of the proposed algorithm to achieve a blind child-mother matching without a priori knowledge of the mother particle corresponding to a given child particle. In the investigation of the effect of initial flaws on the particle breakage, it is found that the initial flaws have a relatively slight influence on the stiffness behavior but affect the particle strength significantly. The particle crushing behavior including the peak crushing load, fracture pattern and the configuration of fragments is subject to the combined effects of particle morphology features and initial flaws. In one-to-one mapping of sand particles, for a given strain rate it is found that, shear modulus and friction coefficient affect the initial stiffness, the peak load and the dilation significantly. The whole framework, and the simulation results within, demonstrated that the proposed modeling approach is capable of reproducing macroscopic (e.g., deviatoric stress response and volumetric response) and particle level (e.g., displacement, rotation and branch vector orientation) behaviors that were very similar to what occurs within the physical experiments, validating the effectiveness of the proposed one-to-one mapping technique and particle fragmentation scheme.